Label Encoding

Label Encoding

What is Label Encoding?

Label encoding is a process of assigning numerical labels to categorical data values. It is a simple and efficient way to convert categorical data into numerical data that can be used for analysis and modelling.

The basic idea of label encoding is to assign a unique integer to each category in a categorical variable. For example, if we have a categorical variable “colour” with categories “red”, “green”, and “blue”, we can assign the labels 0, 1, and 2 respectively. This allows us to represent the data numerically, which is necessary for many machine learning algorithms.

Basic Example of Label Encoding

Suppose you have a dataset of fruits and you want to encode them based on their type:


You can use label encoding to convert the categorical feature “Type” into numerical values. The encoding would look like this:


In this case, we assigned the label “0” to fruits and “1” to vegetables. Now, we can use these numerical values to feed the data into a machine learning algorithm for further analysis.

To further perform label encoding in Python (using the scikit-learn library)

from sklearn.preprocessing import LabelEncoder

# Create a sample dataset
fruits = ['Apple', 'Orange', 'Banana', 'Carrot', 'Tomato', 'Potato']
fruit_types = ['Fruit', 'Fruit', 'Fruit', 'Vegetable', 'Vegetable', 'Vegetable']

# Create a LabelEncoder object
le = LabelEncoder()

# Fit and transform the categorical data
encoded_types = le.fit_transform(fruit_types)

# Print the original and encoded data
print('Original Data:', fruit_types)
print('Encoded Data:', encoded_types)


Original Data: ['Fruit', 'Fruit', 'Fruit', 'Vegetable', 'Vegetable', 'Vegetable']
Encoded Data: [0 0 0 1 1 1]

In this example, we first create a sample dataset consisting of a list of fruits and their corresponding types. Then, we create a LabelEncoder object and fit it to the fruit types data. Finally, we transform the fruit types into encoded values using the fit_transform() method of the LabelEncoder object. The resulting encoded values are printed to the console.

What are the Benefits of Label Encoding?

The benefits of label encoding include:

  • Simplification of data: Label encoding can help simplify data by converting categorical variables into numerical values. This can make it easier to perform statistical analysis and machine learning on the data.
  • Better performance in certain algorithms: Some machine learning algorithms, such as decision trees and random forests, work better with numerical data rather than categorical data. Label encoding can help improve the performance of these algorithms.
  • Reduced memory usage: Numerical data typically takes up less memory than categorical data, which can be useful when working with large datasets.
  • Flexibility: Label encoding can be applied to a wide variety of categorical variables, making it a flexible tool in data preprocessing.
  • Preserves ordinality: If the categorical variable has some natural ordering, label encoding can preserve this ordinality in the resulting numerical values, which can be important in certain types of analysis.

It’s important to note that label encoding may not be appropriate for all types of categorical variables. For example, if there is no natural ordering to the categories, assigning numerical values can be misleading. In those cases, one may consider using one-hot encoding or other methods of categorical encoding.

We label encode because many machine learning algorithms require numerical data as input. By converting categorical data into numerical data, we can use a wider range of machine learning algorithms to model and analyse the data. Additionally, label encoding can be useful in feature engineering, where we transform the original data into a more suitable format for machine learning algorithms. It can also be applied in classification problems, regression problems, recommendation systems, and clustering.

Additional Resources